Recommendation
Generate ranked item recommendations using trained recommendation models
Recommendation inference generates a ranked list of items for users (or similar users/items for a query). Provide user-item interaction data in the same format used during training.
Available Models
- Matrix Factorization (SVD) – Collaborative filtering via singular value decomposition
- Matrix Factorization (sklearn) – Sklearn-based sparse SVD matrix factorization
- Item-Based KNN – Recommend based on item-to-item similarity
- User-Based KNN – Recommend based on user-to-user similarity
- Content-Based TF-IDF – Recommend based on item text/metadata similarity
- Embeddings Similarity – Dense embedding-based item similarity
- BERT4Rec – Sequential recommendation using BERT-style self-attention
- Hybrid CF + CB – Combines collaborative filtering with content-based signals
- Association Rules – Recommend based on frequent item co-occurrence patterns